Machine Learning for Methane Detection and Quantification From Space: A survey

Methane (CH<sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">4</sub>) is a potent anthropogenic greenhouse gas, contributing 86 times more to global warming than carbon dioxide (CO<sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub>) over 20 years, and it also acts as an air pollutant. Given its high radiative forcing potential and relatively short atmospheric lifetime (nine years plus or minus one), CH<sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">4</sub> has important implications for climate change; therefore, cutting CH<sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">4</sub> emissions is crucial for effective climate change mitigation. This article provides an exhaustive list of operational CH<sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">4</sub> point source detection sensors in the short-wave infrared (SWIR) bands. It reviews the state of the art for traditional as well as machine learning (ML) approaches. The architecture and data used in such ML models are discussed separately for CH<sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">4</sub> plume segmentation and emission rate estimation. Traditionally, experts rely on labor-intensive manually adjusted methods for CH<sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">4</sub> detection. However, ML approaches offer greater scalability. Our analysis reveals that ML models outperform traditional methods, particularly those based on convolutional neural networks (CNNs), especially the U-Net and transformer architectures, if enough representative training data are available. These ML models extract valuable information from CH<sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">4</sub>-sensitive spectral data, enabling more accurate detection. Challenges arise when comparing these methods due to variations in data, sensor specifications, and evaluation metrics. To address this, we discuss existing datasets and metrics, providing an overview of available resources and identifying open research problems. Finally, we explore potential future advances in ML, emphasizing approaches for model comparability and large dataset creation; address issues with ML; and provide trends, future research directions, and the European Union’s forthcoming CH<sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">4</sub> strategy.